Experimental setup and procedures
This protocol is extracted from research article:
Hard limits to cognitive flexibility: ants can learn to ignore but not avoid pheromone trails
J Exp Biol, Jun 4, 2021; DOI: 10.1242/jeb.242454

Experiment 1 explored whether ants could learn to avoid pheromone trails after the trails were associated with punishment. We tested 31 ants using a Y-maze (Fig. 1A) and marked its negative arm with a pheromone trail solution, by drawing 6 µl of the solution in an even line over the overlay using a calibrated capillary tube. This produces a strong but realistic pheromone trial, and elicits trail following indistinguishable from natural trail following (von Thienen et al., 2014). The positive arm was covered with a dichloromethane trail (6 µl) as a solvent control. A droplet of bitter tasting quinine solution (60 mmol l−1; Avarguès-Weber et al., 2010) was placed at the end of the negative arm and a droplet of 1 mol l−1 sucrose solution at the end of the positive arm of the Y-maze.

Experimental setup. (A) The Y-maze used in experiments 1, 2 and 4 and in the unrewarded learning tests in experiment 3. The arms were 10 cm long and 1 cm wide. The ant is shown to scale. The specific setup shown is experiment 2, where ants were punished on the pheromone-marked arm using both an electric shock (see B) and quinine solution. Experiment 1 was the same as experiment 2, but without the electric shock device (‘shocker’). In experiment 4, the unscented arm was rewarded, and instead of pheromone the punished arm was scented with lemon. The sides of the positive and negative arms (left/right) were pseudo-randomized between trials. (B) Schematic diagram of the shocker. An attachment part was affixed to the bottom of the path, and the shocker attached via magnets to allow rapid removal for cleaning and paper overlay replacement. The shocker was 3D printed from PLA (see Materials and Methods).

The pheromone solution was created by freeze-killing workers and dissecting out their hindguts – the glandular source of the L. niger pheromone trail (Bestmann et al., 1992; von Thienen et al., 2014). Four glands were macerated in 2 ml dichloromethane. The pheromone solution was separated into 1 ml aliquots and stored at −20°C between experiments. During the experiment, the aliquot used was kept on ice. Dichloromethane and pheromone trails were applied to the paper layers just before placing them on the Y-maze arms. Between trials, all paper layers were replaced to remove any pheromones left by the focal ant during the previous trial.

To start the experiment, we touched a small piece of paper to the nest floor. The first ant to climb onto the paper was put on the starting point of the Y-maze, which had the positive arm covered with the control solvent, and the negative arm covered with a pheromone trail. After reaching the sucrose drop on the positive arm, the ant was marked with acrylic paint on its abdomen while drinking the sucrose solution. Afterwards, she was allowed back to the nest. After unloading the sugar, the ant was brought back onto the Y-maze for the next trial. In each of the following 25 trials, the ant's initial choice (left or right) was recorded when the ant crossed a line 1 cm from the bifurcation, while the final choice was recorded when the ant reached one of the droplets (quinine on the negative arm, sucrose solution on the positive arm). As ants ran on the runway, we also measured the time the ants took from entering the runway to reaching the sucrose solution on the positive arm.

The positive and negative arms and their respective reinforcement and punishment were switched between trials, following a pseudo-randomized order [L–R–L–R–R–L–L–R–R–R–L–R–L–L–R–R–L–L–L–R–R–L–L–R–R and its reversed sequence, where L (left) and R (right) indicate the arm of the maze containing the sucrose reward] to ensure the subjects would not associate the reward and punishment with any particular side.

During experiment 1, we noticed that ants very rapidly learned to carefully probe the droplet with their antennae before drinking. This enabled them to avoid punishment for an incorrect choice, thus greatly reducing the cost of errors. Such reduced error costs may not have been sufficient to promote learning. We thus designed a further experiment (experiment 2) in which we introduced an electric shock device (henceforth ‘shocker’; Fig. 1B) as an unavoidable punishment. Shockers were affixed 4 cm from the bifurcation on each of the Y-arms. They consisted of a 3D-printed PLA body (an STL file can be downloaded in modifiable form from https://www.tinkercad.com/things/dl5ce7taHaI-ant-zapper), which offered a tunnel narrowing from 1 cm (the width of the Y-maze arms) to a 2 mm gap (Fig. 1B). The floor of the gap was covered with two slightly disconnected copper plates (Fig. 1B, left), which were connected via wires to a button and a laboratory power supply. When the ant walked through the gap, thereby touching each copper plate with at least one of her legs, she closed the electric circuit (if the shocker was activated) and got shocked with 7.5 V (Roussel et al., 2012). The front and back of the shocker were equipped with a polytetrafluoroethylene (Fluon®) plastic plate, preventing ants from passing the barrier except via the tunnel. Apart from adding the shocker, experiment 2 was identical in design to experiment 1 above. A total of 31 ants were tested.

While the overall procedure of experiment 2 was identical to that of experiment 1 above, we added two methodological improvements. Firstly, we established two pre-training trials prior to the trials, ensuring a standard baseline experience across subjects. Secondly, we added two unrewarded learning tests after trial 20 and 25, to increase the chance of uncovering any cryptic learning that may have occurred (Bortot et al., 2019).

In the pre-training trials, we confronted the ant with one rewarded trial followed by one punished trial, both presented on a linear runway (21 cm long, 1 cm wide). In the first trial, paper overlays covered with a control solvent were placed on the runway, which was also equipped with an inactive shocker and a droplet of sucrose solution that was presented behind the device. After reaching the sucrose drop, the ant was marked with acrylic paint on its abdomen while drinking the sucrose solution. Afterwards, she was allowed back to the nest. After unloading the sugar, the ant was allowed back onto the linear runway for a second pre-training trial. In this second trial, the ant was confronted with a linear runway covered with a pheromone trail, an activated shocker, and a droplet of quinine solution behind the device. After the ant experienced both negative stimuli (shock and quinine), it was transferred to the Y-maze for the test. Within the pre-training trials, the punishment trial was always carried out after the reinforcement trial to ensure the ant's participation.

During unrewarded learning tests, the arms of the Y-maze were equipped with the respective paper layers (control solvent or pheromone trail), but both shockers were inactivated, and the sucrose and quinine droplets were replaced by neutral water droplets. After entering one arm of the Y-maze, the connection to the Y-maze stem was interrupted for 1 min, therefore only allowing the exploration of both Y-maze arms. We recorded the overall time the subject spent on the correct (positive) arm before replacing the water droplets with the respective liquids (sucrose or quinine solution), allowing the ant to drink from the sucrose, and to return freely to the nest for further testing.

Ants in experiments 1 and 2 developed a strong but mostly arbitrary side bias (see Results). We were concerned that the ability to develop a favoured side may interfere with the task of learning to avoid the pheromone trail. We thus developed experiment 3, which excluded this possibility by employing a go/no-go paradigm on a linear runway. Twenty-nine ants were tested, using a 21 cm linear runway that was either marked with a pheromone trail (12 μl) and offered a droplet of bitter tasting quinine solution at the end (punishment trial) or marked with a dichloromethane trail (12 µl) and contained a droplet of 1 mol l−1 sucrose solution at the end (reward trial). As ants ran on the runway, we measured the time the ants took from entering the runway to reaching the droplet at the end as well as the number of U-turns they performed on their way to the droplet. A U-turn was defined as turning around and moving at least 1 cm towards the nest and away from the droplet. In addition, we carried out two unrewarded learning tests on a Y-maze, as described in experiment 2, and recorded the ant's choices (left or right). All trials were video-recorded, and videos were subsequently analysed by a naive coder, to ensure blindness to trial type (punishment or reward trial).

To start the experiment, we touched a small piece of paper to the nest floor. The first ant that climbed onto the paper was put on the starting point of the linear runway. The first trial was always a reward, no-pheromone trial to ensure the ant's participation. A total of 25 trials were presented in a pseudo-randomized order (R–P–R–P–P–R–R–P–P–P–R–P–R–R–P–P–R–R–R–P–P–R–R–P–P, where R and P are the reward and punishment, respectively). When reaching the sucrose solution in the first reinforcement trial, the ant was marked with acrylic paint on its abdomen while drinking. Afterwards, the marked ant was brought back to the nest, where she could unload her sucrose load, and was then brought back to the start of the runway for a punishment trial with pheromone. After finding and tasting the quinine solution in the punishment trials, the ant was delayed for 30 s (the time it approximately took her to drink the sucrose solution in the reward trials), before she was brought back to the nest. After trial 10 and 20, unrewarded learning tests on a Y-maze were added, as in experiment 2.

The aim of experiment 4 was to show that it is within the ants’ capacity to learn to avoid a neutral chemical signal (here: lemon odour) after it was associated with a punishment. We tested 12 ants using the same setup as in experiment 2 (Fig. 1) but replaced the pheromone trail on the negative arm of the Y-maze by lemon scent while the positive arm was covered with unscented paper overlays. Lemon-scented paper overlays were produced by placing unscented paper layers in an airtight container with lemon essential oil for at least 24 h; these were taken out of the container just before applying them to the runway. Between trials, all paper overlays (scented and unscented) were replaced to avoid orientation via pheromones left during the previous trial.

Five ants were allowed to walk up a bridge, leading to a Y-maze, one arm of which was covered with an unscented paper overlay, and the other covered with lemon-scented paper. The first ant choosing the lemon-scented arm first was selected to be the subject, while ants choosing the unscented arm of the Y-maze were gently brought back into the nest. This was done in order to ensure that no ants with an innate aversion to lemon odour were tested, thus making the experiment very conservative. The pre-selected subject then carried out two pre-training trials, 12 trials [following a pseudo-randomized order: L–R–L–R–R–L–L–R–L–R–L–R and its reversed sequence, where L (left) and R (right) indicate the arm of the maze containing the sucrose reward] and two unrewarded learning tests after trial 8 and 12, all as described in experiment 2, but with lemon scent instead of a pheromone trail.

Note: The content above has been extracted from a research article, so it may not display correctly.



Q&A
Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.



We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.